<p>Detecting a life-threatening atherosclerotic plaque, called Thin Cap Fibroatheroma (TCFA) or vulnerable plaque in Virtual Histology-Intravascular Ultrasound (VH-IVUS) images is a challenging task. To improve the reliability of detecting TCFA early, a new segmentation method, namely the Plaque Burden Assessment by Local Search (PBALS) algorithm, has been proposed using VH-IVUS images, and geometric features have been extracted. Additionally, a hybrid feature extraction method uses discrete cosine transform (DCT) and discrete wavelet transform (DWT) algorithms proposed to extract the texture feature. These features have been used by a set of proposed ensemble classifications to detect the non-TCFA from TCFA plaques with expected reliability and robustness. 566 in-vivo IVUS images and their matching VH-IVUS images gathered from 10 patients were used in the experiment. Based on the results, the combination of VH-IVUS and IVUS features performs better than standalone VH-IVUS features. Moreover, our proposed methods outperformed most other state-of-the-art methods regarding accuracy, specificity, and sensitivity. Some of our proposed ensemble methods showed promising results even with a few features. However, more VH-IVUS and IVUS data are needed to use the proposed methods as an aid to cardiologists in identifying the types of plaques and assessing their vulnerability.</p>